STORM TIDE RISK ASSESSMENTS IN TROPICAL AND SUB TROPICAL AREAS INCLUDING CONSIDERATION OF CLIMATE CHANGE IMPACTS AND EMERGENCY MANAGEMENT NEEDS

Ross Fryar1, Dr Bruce Harper2,

1 Manager Water and Environment, GHD, Brisbane

2 Managing Director, Systems Engineering Pty Ltd, Brisbane.

Abstract The assessment of related storm tide risks is an essential planning tool for coastal councils in northern Australia, made even more critical due to the possible impacts of sea level rise and the climate change modification of storm intensity and frequency. With rapidly increasing coastal based centres of population, emergency managers are also under pressure and in need of better information and advice on how to cope with potential large scale disasters. This paper outlines key issues and approaches in dealing with the potential impacts of storm tide through reference to a number of recent studies in (in particular the Whitsunday, , South East Queensland and Innisfail regions). For each of these locations, climate change considerations have been fully built-in to storm tide study methodologies providing the opportunity to also service emergency management planning and training needs. The studies are based on the best practice methodology recommendations from the Queensland Climate Change Study conducted 2001 – 2004 . Aspects of these studies are presented, focussing on a blend of method and function, including best practice methodology, the importance of model calibration and testing, preparing results for emergency and planning functions, and the provision of insights into the specific likely climate change impacts for each region.

Key Words: climate change, sea level rise, tropical cyclones, storm tide, emergency management also considered, the risk to Queensland Introduction coastal communities is potentially Tropical cyclones generating storm tides and significantly increased, with areas of cities flooding in Australia rank on top of the list of such as Cairns and the Gold Coast having a extreme events (geohazards), with large number of low lying properties. Key Queensland, Western Australia and the infrastructure (water, power, airports) will also particularly exposed. When be subjected to higher risk, with implications population is taken into account, Queensland for emergency management. coastal communities account for a high proportion of the population at risk from such events. Despite the existence of this risk, the Storm Tide and Impacts theory and application of storm tide The definition of and storm tide predictions are generally poorly understood is now generally well understood by by those most likely to have a need to apply emergency management and planning the results. The need for informed studies to personnel, as is the range of potential risks to be undertaken is therefore of paramount population and infrastructure as indicated importance. above. However, the process of estimating storm tides, and how to best make use of When sea level rises associated with climate such studies is not understood to the same change and increasing cyclone intensity are degree. Extreme storm tide levels caused by tropical Assemble cyclones cannot be estimated solely on the Data basis of historically measured water levels.

This is because the available record of Historical Cyclone Coastal Tide Coastal Storm Tide Occurrence Bathymetry Information Infrastructure tropical cyclones affecting any single location Data on the coast is quite small, the resulting

Determine storm surge response is often complex and Establish Regional Calibrate Wind Hydrodynamic Cyclone against Althea Information very site specific, and the final storm tide is Models dependent on the relative phasing with the Climatology astronomical tide. Hence, measured storm Define Wide Establish Range of Spectral Wave Cyclone tide data alone is typically inadequate for Model extrapolation to very low probabilities of Parameters occurrence. Simulate Tropical Storm Surge Cyclone Wind and Wave and Pressure To overcome this problem, it is necessary to Responses Model formulate a statistical model of the coastal Parameterise Simulate and Construct Validate Assess Numerical Predict Statistical Statistical Regional region that will attempt to re-create the Model Return Model Model Impacts observed region-wide tropical cyclone Responses Periods climatology and numerically generate long sequences of potential storm tide scenarios, a process that has been applied to many Figure 1 Example coastal hazard study storm tide studies along the Queensland methodology. coast (for example, the Whitsunday, Townsville and Innisfail regions) and throughout Australia and nearby regions Planning and Emergency Management (Harper at al. 2007). One of the most commonly encountered queries when commencing a storm tide study There remain two key aspects which are not is that relating to how emergency always consistently addressed. These are: management procedures might be affected, (a) understanding the relative importance of and whether the study will deliver a better planning (as compared to emergency understanding of “exact” water levels for any management), and (b) the inclusion of the given cyclone. This line of thought fails to potential affects of climate change (other than recognise that during an extreme event, through a simplistic adjustment of sea level). emergency management procedures are Each issue is addressed in the following reliant on an inherently uncertain forecast of sections with an overview of the common peak water level provided by the Bureau of analysis methodology is provided in Figure 1. Meteorology, and that the extent of areas that may require evacuation for a given storm tide This approach is fully consistent with the level can be easily defined when needed. recommendations from the Government- sponsored Queensland Climate Change and Hence, it is not always apparent that a Community Vulnerability to Tropical Cyclone fundamental purpose of conducting storm – Ocean Hazards Assessment-Stage 1 tide studies should be to facilitate future (Harper 2001). This study (hereafter referred planning. In this context, planning primarily to as the “Blue Book”) provides a state-of-the- refers to land use planning (i.e. the reduction art methodology for ocean hazard of risk to life and property through a reduced assessments and a technical pathway to best encounter probability in the form of practice assessment of climate change appropriate siting). A well conducted study impacts. Building on the advantages of the will therefore provide two benefits: provision QCC methodology, GHD teamed with SEA in of a greater level of the understanding of risk, 2007 to provide a revised “Blue Book” such that better informed decisions can then standard of service to coastal councils in be made in both a planning and emergency order to meet their regional study context. requirements. Addressing Climate Change Details from a number of completed studies Climate change can exert an influence on follow. The possible effects of greenhouse- cyclonic activity, and hence storm tide in induced climate change have been several ways. It can generally be assumed considered in these studies, whereby a that: possible future climate scenario is simulated and those results compared with the estimates for “present climate”.  A rise in sea level will occur;

 A rise in mean sea level (MSL) will also lead to a rise in HAT and the tidal Whitsunday Region characteristics may also change slightly as The Whitsunday Storm Tide Study, a result, but this effect is commonly completed in 2003, was the first to apply ignored. Also, although AHD is based on “Blue Book” regional standards of storm tide MSL, it is assumed that the AHD datum risk analysis to a Local Government fine will remain where it is now. scale study. A principal challenge in the area was the complex nearshore island  An increase in tropical cyclone maximum environment and the very high tidal ranges, potential intensity or MPI1 will also occur. which vary markedly throughout the region It should be noted however, that it is not a (refer Figure 3). straightforward concept to apply to the statistical description of individual cyclone central pressure values. The interpretation made here is that the most intense of cyclones may increase their intensity but that not all cyclones will be more intense.

 The way that this can be applied is illustrated in Figure 2, whereby the potential % increase (relative to 'p) is blended into the present climate description used by the statistical model.

990 Present Climate 970 +10% MPI +20% MPI 950

930

910

890

Central Pressure pc (hPa) Pressure Central Figure 3 The Whitsunday region study area 870 with its complex island environment and high 850 tidal variability 1 10 100 1000 10000 Return Period (y) The study highlighted the principal locality at Figure 2 Assumed Possible Changes in the risk was the Repulse Bay area near the Intensity of Tropical Cyclones under mouth of the Proserpine River, where the low Enhanced Greenhouse Conditions lying land and limited evacuation routes combined with a dynamic storm surge amplification effect to produce increased risk. 1 The MPI is the theoretical peak storm intensity A secondary benefit was the increased level available from the regional thermodynamic of understanding that the study provided: properties of the atmosphere and ocean, subject storm tide levels can be highly variable along to a range of other favourable conditions. the coastline for any given event (whether a cyclone or a defined recurrence interval such as the 1 in 100 year storm tide level).

To assist Council in their emergency planning, a series of specific cyclone scenarios were also modelled, impacting on different parts of the region. This produced results that had been unexpected by the emergency managers; namely that locations some considerable distance from the cyclone landfall position could be more adversely affected than those at the point of landfall.

Townsville and Thuringowa Region The Townsville-Thuringowa Storm Tide Study was completed under the auspices of the Federal Government’s Natural Disaster Risk Figure 4. Large scale grids utilised for Mitigation Program (NDRMP). The aim of Townsville study this program was to minimise / reduce future costs associated with the occurrence of Key features of the analytical process to natural disasters through better planning determine storm surge (and hence storm tide processes. For this study, the focus was on levels) for each of the nominated design inundation associated with a range of events included: statistical storm tide events, and the levels of risk and exposure associated with predicted  Calibration to (1971) inundation.  Verification against Cyclone Aivu (1989) As with the Whitsunday project, the model  Completion of sensitivity analysis model development process utilised four different runs software packages which addressed:  Modelling of over 320 different cyclones,  Tropical cyclone wind and pressure approaching on a number of different

 Hydrodynamics tracks.

 A spectral wave analysis  Simulation of a 50,000 year period (90,000 cyclones) using a statistical model.  Statistical consideration of surge and tide predictions.  Probabilistic tide and wind verification. These were developed using a system of Results were presented in terms of predicted nested grids. The largest of these grids (refer surge height, wave set-up height and finally Figure 4), which provides data as to the sea- the resultant storm tide level for events bed and hence water depth, extends almost ranging from the 50 year average recurrence 800 km off-shore, and almost 1400 km along interval (ARI) up to the 10,000 year ARI the coastline. The smallest grid is 30km x event. 35km, with a grid size of 55m. A vulnerability assessment was then undertaken. This was based on a review of the predicted and mapped storm tide levels, and the implications for critical infrastructure, as identified in consultation with Townsville and Thuringowa Councils. Tables were produced which detailed items of infrastructure potentially affected. In addition, a review of the Counter Disaster Plan was undertaken, with particular attention to Enhancing Emergency Training and evacuation routes. Awareness

It was estimated that for the 100yr event, As part of the study, Townsville City requested over 1000 habitable properties would be provision of a storm tide warning system. The subject to some form of inundation. The resulting SEAtide model has the ability to population at risk was estimated at readily simulate many hundreds of possible approximately three times the number of tropical cyclone scenarios on the basis of a properties, with depths of greater than 0.5m “best estimate” forecast provided by the associated with a higher risk of death. For Bureau of Meteorology. Uncertainty in the the 10,000 year event, the number of properties affected would be significantly forecast is used to produce a probability higher, with most of these subject to depths spread of various levels of inundation of 0.5m or more. occurring rather than relying on a single estimate with unknown precision. As well as providing warning information, SEAtide can Impact of Climate Change also applied as a realistic training tool, The climate change scenarios considered in enabling emergency management staff to the Townsville study are summarised in better understand the dynamic effects that Table 1. can occur during extreme tropical cyclones. An example of a SEAtide output is given in Figure 5 below. This indicates which region Table 1 Enhanced Greenhouse Scenarios is most likely to be impacted as the cyclone Year Increase Increase in Increase approaches, and the magnitude of storm tide in Maximum in Mean predicted. frequency Potential Sea Level Intensity

% % m

2050 10 10 0.5

2100 10 20 0.9

These scenarios were simulated in the storm tide modelling packages. Results for the year 2050 showed an average increase in storm tide level across all sites of 0.5 m at both the 500 yr and 1000 yr return period and 0.7 m for the 10,000 yr. The corresponding averaged results for 2100 were higher again, with increases compared to the current time of the order of 1.1 m, 1.2 m and 1.7 m Figure 5 Example SEATide prediction model respectively. interface for the Townsville region in North Queensland A similar approach has been adopted for the ongoing storm tide study for the Cassowary Coast region, which was subjected to Unlike a numerical modelling system in 2006. requiring specialist long term staff training and support, SEAtide provides a user friendly interface that requires little specific training or maintenance. This approach not only allows training and a raised awareness of storm tide (from an emergency management perspective), but can also be applied to non- Storm Tide Studies in South East Qld cyclonic storm tide, coastal wind and wave Over the past 4 to 5 years, a number of storm threats. tide studies have been conducted for Queensland coastal councils. The Revisiting Best Practice Methodology complexity of storm tide has been dealt with in a different manner for many of these The SEQ Disaster Management Group studies. Whilst the QCC study (Harper 2001) (SEQDMAG) commissioned a study in 2007 “Blue Book”, was commissioned to provide to review a range of completed regional some consistency of approach, studies were storm tide risk reports, and to compare the delivered to different standards. Further, standard of each report in relation to the while applicable to SEQ, the “Blue Book” outcomes of the first part of the study, namely methodologies had been developed for a review and update of recommended best Queensland regions not on the “fringe” but at practice methodology for assessing storm the “centre” of cyclone related impacts. tide in the South East Queensland (SEQ) region. Several factors therefore drove the need to review best practice methodology and to The need for the study was driven in part by review completed studies against this the recognition of increasing risk, and the methodology. These consisted of: a possible recommendations of earlier studies. One of change to cyclone behaviour; a possible these, completed in 2001 by Geoscience increase to SEQ storm tide risk; a need for Australia (GA), was the AGSO Cities Project regionally consistent outcomes (results), and in South East Queensland (Grainger and the current lack of a regionally specific Hayne 2001). This earlier work examined the approach created a need to review the status impact of natural hazards and the risk they and approach of the SEQ region to storm tide posed on the communities from Caboolture hazard. Shire to the Gold Coast City Council including Ipswich City Council. Storm tide was Findings identified in that study as a natural hazard of some significance, due to both tropical and The total score across all of the six reviewed extra-tropical storm systems. SEQ study reports was less than 50% of the hypothetical perfect score from a combined It is clear that the region faces significant planning and emergency response demographic changes, with an ever perspective. This reflected the fact that increasing demand for development of all several of the studies did not have a specific kind, including residential, to be sited emergency response scope and some had adjacent to the coastline. With more people no atmospheric or limited hydrodynamic moving to the coastal zone and evidence of modelling scope. climate change, the level of risk may rise. In this case, climate change effects are defined Whilst the QCC study itself scored high in not only in terms of increasing water levels, climatology, atmospheric modelling, statistical but also an increase in the frequency of modelling and documentation, no wave incidence of cyclones, as described below. modelling south of the Sunshine Coast nor non-cyclonic events were considered in SEQ. Under current climatic conditions, South East Queensland lies on the southern fringe of For each of the location specific studies, exposure to the most intense impacts of climatological and atmospheric modelling tropical cyclone. Uncertainties surrounding elements typically scored poorly, mainly the effect of likely climate change on cyclone owing to a lack of disclosure of the behaviour have raised questions of potential methodology. For example, it is considered increase in risk to the SEQ region due to critical that the forcing applied to the severe storm and cyclone related impacts. hydrodynamic modelling is representative and unbiased and that the selection of modelled events is well founded on a comprehensive storm climatology that acknowledges the need to identify different In terms of wind strength, the intensity of storm populations and their associated Larry at landfall was the subject of some intensity and scale distributions. debate between the meteorological, coastal and wind engineering communities. During In terms of the possible staging of such the event, Larry was forecast to be a improvements, the need to review the Category 5 storm at landfall, with the historical storm data (intensity and scale) and potential for significant and widespread develop representative regional storm destruction. After the event, damage surveys climatologies was identified as a major critical confirmed that the peak gust wind speeds path item that is common to all LGA regions. experienced near ground level in standard exposure conditions were most likely in the Finally, the need was identified for the range of 55 to 65 m s-1 (200 to 235 km h-1). development of improved atmospheric Combined with the reliably measured central models of each storm type (TC, East Coast pressure of 955 hPa, this placed Larry Low etc) and the adoption of responsible potentially in the mid to high Category 3 class Greenhouse-sensitive perturbations. only.

However, the seemingly high beach debris Cyclone Larry and Innisfail levels in some areas, especially near the Tropical Cyclone Larry (Innisfail, March 2006) landfall, generated conjecture within the was the most significant event in the Bureau that the storm was “unusual” and may Queensland region for the past 20 years. Its have delivered unexpectedly high surface impacts included moderate to severe wind winds. Nevertheless, officially Larry remains damage to buildings in the Babinda-Tully a Category 4 storm immediately prior to coastal region while storm tide effects were landfall and offshore wind and pressure limited to below the threshold of potential calibrations conducted by SEA were devastation, mainly due to the coincidence consistent with that assessment. with neap tides (Boswood and Mohoupt 2007). The peak measured surge magnitudes Because the storm crossed the Great Barrier were 2.3 m at Clump Point Jetty, with a total Reef lagoon it demanded hydrodynamic stillwater level of 2.6 m AHD, and 1.3 m at modelling at a scale sufficient to resolve the Mourilyan Harbour, being 1.6 m AHD. The major reef passages. This was economically peak surge was estimated to have occurred achieved using a nested regular grid in between these two gauges, which are Delft3d, down to 550 m. Spectral wave approximately 30 km apart. modelling was also performed by SEA to the same resolution. The accurate modelling of Consideration of this complex event has astronomical tides over the period of neaps provided an opportunity to further test and when the storm occurred proved difficult due develop the degree of wind, wave and surge to small errors in the offshore tidal accuracy that can be obtained through the boundaries. Accordingly, predicted tides were adoption of a robust methodology. The used to form a total storm tide time history. ongoing work described below is being conducted for Johnstone Shire Council (now Results of the study are summarised in the Cassowary Coast Regional Council). Figure 6, whereby it proved possible to almost exactly reproduce the previously The highest astronomical tide (HAT) in the thought “unusual” surveyed beach levels. The Innsifail region is around 2 m AHD. This fact, various storm tide components of tide, surge, combined with the rapid movement of Larry wave setup and 1% wave runup all proved (32 km h-1), meant that beach erosion reasonably consistent and reflected the impacts were relatively minor. With wave complex interaction between the storm track, effects not measured, the only available data the wind and pressure fields and the array of relating to intermediate storm tide estimates reef passages. was that from debris or inundation marks (surveyed up to 5.2 m AHD). This work again highlights the importance of building confidence in modelling methodologies so as to form a firm basis for coastal centres. The tools now exist to help future climate change sensitivity testing. The decision makers account for these pressures. full study is currently being completed.

6.0 References Bingil Bay Debris Inundation 5.0 Tide Gauge Boswood, P.K. and Mohoupt, J. (2007) Etty Bay Kurrimine Beach Modelled Debris Tropical Cyclone Larry: Post Cyclone Coastal Clump Point Modelled Storm Tide 4.0 Cowley Beach Field Investigation, Coastal Sciences Flying Fish Point Modelled Tide+Surge Mission Beach South Mission technical report, Env. Protection Agency Wongaling Beach 3.0 Tully Heads Granger K and Hayne M (Eds.) (2001) Cardwell Clump Point

Water Level m AHD m Level Water Natural hazards and the risks they pose to 2.0 South-East Queensland. Produced by Lucinda

Mourilyan AGSO- Geoscience Australia in conjunction 1.0 with the Bureau of Meteorology and SEQ Local Government Organisations. 0.0 -17.4 -17.6 -17.8 -18 -18.2 -18.4 -18.6 Commonwealth of Australia. Latitude GHD and SEA (2007a) Townsville – Figure 6 Summary comparison of alongshore Thuringowa Storm Tide Study. Prepared for modelled and measured peak water levels Townsville and Thuringowa City Councils during TC Larry GHD and SEA (2007b) South East Queensland Storm Tide Review. Discussion and Conclusions Recommendations for Modelling Risk Storm tides occur as a response to severe Assessment and Mitigation Strategies. meteorological condition. In Queensland, this Prepared for Caboolture Shire Council (on is most commonly seen to be a consequence behalf of the South East Queensland of cyclonic activity, though other low pressure Disaster Management Group). systems can also generate storm tide. GHD and SEA (2003) Whitsunday Storm As population demands imposed on coastal Tide Study. Prepared for Whitsunday Shire regions continue to grow, the inherent level of Council (now Whitsunday Regional Council) risk has also grown. This is exacerbated by climate change risk, whereby higher sea Harper BA (Ed.) (2001) Queensland climate levels and more intense (i.e. lower pressure) change and community vulnerability to cyclones may occur. tropical cyclones: ocean hazards assessment - stage 1. Report prepared by Systems This paper has provided an overview of the Engineering Australia Pty Ltd in conjunction assessment of storm tide risk through with James Cook University Marine Modelling reference to a number of studies completed Unit, Queensland Government, March, over the past five years. The paper has 382pp. addressed in broad terms: the importance of appropriate methodology, the higher level of Harper B, Hardy T, Mason L, Fryar, R, (2007) understanding that results, and the Developments in Storm Tide Modelling and implications of climate change on each of Risk Assessment in the Australian these. Furthermore, the benefits that this Region..Proc. WMO/IOC JCOMM 1st provides to those responsible for planning Scientific and Technical Symposium on and emergency management is significant. Storm Surges, Seoul, Korea, 2-6 Oct.

Understanding the potential impacts of climate change and sea level rise needs to Ackowledgements be a key factor in coping with demand on The authors would like to acknowledge each of our clients, and the efforts of Dr Ivan Botev